1

The Development of Global Coherence in Life Narratives across Adolescence:

Temporal, Causal and Thematic Aspects

Tilmann Habermas and Cybèle de Silveira

Goethe University of Frankfurt am Main, Germany

Preprint version

In press with Developmental Psychology, May 2008

Acknowledgments. This article is based on data from a study on the development of life

narratives which was supported by the German Research Council (DFG), grant #HA 2077. We

thank the children and adolescents, teachers and directors of the Comenius Grundschule and the

Bettina Gymnasium, Frankfurt, for their participation and kind support. Thanks also go to

Verena Diel and Martha Havenith, who collected the data with the second author, for data entry

and coding, which was also done by Andreas Fröhlich, Helge Hachenberg, Michael Schreiner,

Claudius Stauffenberg, Fiona Bonzelius, Leonie Koske, and Christina Herr. We thank David

Rubin and Susan Bluck for discussions of the original ideas, David Rubin, Verena Diel, Stephan

Bongard and Anna Kenney for helpful suggestions on the manuscript, and Paul B. Baltes,

Sigrun-Heide Filipp, and Ursula Staudinger for encouragement and support of this line of

research.

Address: Please send correspondence to the Department of Psychology, Goethe University, PO

Box 111932, 60054 Frankfurt a.M., Germany, or via e-mail to psychoanalyse@psych.uni-

frankfurt.de. Fax +49 69 798 28584, Phone +49 69 798 22541

2

Abstract

Extending the study of autobiographical narratives to entire life narratives, we tested the

emergence of globally coherent life narratives in adolescence, as hypothesized by McAdams

(1985). A total of 102 8-, 12-, 16- and 20-year-olds narrated their lives twice. Half of each age

group participated in tasks designed to train autobiographical reasoning between narrations, the

other half participated in control tasks. Coherence was measured by the relative frequency of

local temporal, causal, and thematic linguistic indicators identified qualitatively at the level of

propositions, as well as by quantitative global rating scales measuring the impressions of the

listeners. Coherence increased across the age span. Overall, repeated narrating and training did

not increase coherence. Crystallized and fluid intelligence, number of negative life events, and

frequency of biographical practices and confiding in others did not contribute substantially to the

prediction of coherence beyond age. Results are interpreted in the context of adolescent identity

development.

Key words: life narratives, adolescence, autobiographical reasoning, life story, autobiographical

memory

3

The Development of Global Coherence in Life Narratives across Adolescence:

Temporal, Causal and Thematic Aspects

Based on early memory abilities such as deferred imitation and recognition (Bauer, 2006)

autobiographical remembering and a corresponding autobiographical self emerge with the ability

to verbally refer to past episodes as experienced by the self, which, by the end of the preschool

years, are organized by the acquisition of the story structure (Nelson & Fivush, 2004). The life

story is a coherent organization of multiple single narratives which are interweaved within the

context of the personal development of the individual. The study presented here sets out to

demonstrate that the development of the organization of narrative remembering continues

beyond childhood. We present the first attempt to systematically test McAdams’ (1985)

hypothesis that the life story develops first in adolescence. We set out by delineating the

theoretical background, then the developmental context of the emergence of the life story.

Finally we introduce the conceptual framework and research strategy pursued in this study.

Identity and the Life Story

The term life story can be used in an objective sense to designate the course of a life as seen

from an observer’s point of view, in terms of what really has happened in a life. Sociologists and

anthropologists are interested in this aspect of lives, which is often termed life history. We will

here reserve the term life story for the subjective representation of the course of a life, as it is

retrospectively remembered, reflected upon, or narrated.

Life stories are used in various contexts in everyday life. They are used to identify

individuals, such as in personnel files or criminal files, and they are used to introduce oneself to

others, both when getting to know someone more intimately and when applying for a job.

Biographies are written about exemplary lives, autobiographies are written to publicize

4

interesting facts about oneself and to explain and justify how one has become the person one is at

present.

In psychology, the intimate link between subjective accounts of one’s own development and

personality has been used by clinicians (Freud, 1905) and personality psychologists (Allport,

1942; McAdams & Pals, 2006) as a rationale for using autobiographical material as a source of

detailed information about an individual. But it was not until Erikson (1968) that the

development of a subjective life story was made the hallmark of the attainment of a mature

psychosocial identity, occurring typically by the end of adolescence. Erikson created the concept

of psychosocial identity to integrate social, clinical, and developmental aspects of identity. It

defines a person’s individuality, consistency across situations, and continuity across time,

requiring the creation of coherence between different role demands and identifications across

situations and across development, as well as the synthesis of social, moral and hedonistic

demands. A well-integrated identity provides a feeling of well-being, a background sense of self-

sameness and continuity, a sense of being at home in one’s body, a sense of purpose and

direction, and an anticipation of being confirmed by significant others (Erikson, 1968).

Erikson’s specific developmental hypothesis is that in modern societies, a new, complex and

flexible form of identity, ego-identity, has to be achieved in adolescence. Adolescence requires

going beyond childhood identifications by consciously confirming or repudiating identifications,

and creating a self-chosen and unique synthesis of commitments to values and roles. Erikson

implies that ego-identity needs to be organized not only synchronously, based on identifications

with roles, but first of all diachronically as a life story. Only a self that is defined by its history

can include the highly individual choices made in adolescent identity exploration, and will prove

more flexible when adapting to continuing change throughout adulthood.

5

In more recent theorizing, the continuing malleability of identity across adulthood has been

stressed (Gregg, 2006; Kroger, 2007). Although Erikson underscored the importance of enduring

commitments formed in adolescence, the flexible form of the life story is especially well-suited

to integrate change across adulthood (Pals, 2006). That identity development implies a certain

stability, however, is evidenced by the so-called reminiscence bump of autobiographical

memories, which describes a relative oversampling of memories from adolescence relative to

other times of life (Rubin, Rahhal, & Poon, 1998). In the case of migrants the oversampled

period is shifted to the time in life in which they migrated (Schrauf & Rubin, 2001),

underscoring that periods formative of identity, most typically adolescence, do leave a stable

trace.

Whereas Erikson never defined the life story as a theoretical concept, McAdams (1985)

explicitly spoke of the life story as a narrative, theorizing that identity takes on the form of a life

story in adolescence. McAdams proposes that ego identity manifests itself in the “good” form of

a life narrative, in its coherence and consistency. McAdams’ thesis does not mean that after

adolescence identity does not change anymore, but only that it is then that it first takes on the

form of a life story, and it is from then on that it is subjected to norms of global textual

coherence. Coherence concerns relations between elements of a text at one point in time and is

not to be confounded with notions of stability across time.

McAdams’ claim that a coherent life story develops first in adolescence has, however, never

been tested. Empirical research on psychosocial identity and the life story has been hampered by

the abstractness and complexity of both concepts. The past 15 years have seen the development

of a rich field of research into narrative identity, which, again following McAdams (1985), has

used autobiographical memory narratives of single episodes (cf. Singer, 2004). But McAdam’s

6

interest in the life story is directed towards differences between persons and not so much towards

development. In keeping with this personological approach, these studies use single memories as

typical examples of the life story that reflect the person’s individuality (e.g., McAdams et al.,

2006), but have not studied the specific form of a coherent life story in adolescence.

The life story may be considered the most comprehensive form of describing an individual.

The life story also serves to create a sense of coherence, unity and purpose, which is considered

to be of prime importance for mental health and well-being (Antonovsky, 1985; Kernberg,

1984). Therefore the development of the life story merits study.

Developmental Precursors of the Life Story

When the ability to construct coherent life narratives emerges, two previously independent

developmental lines merge: the development of narrative and the development of the person

concept. The life story adds a diachronic dimension to personality and contextualizes past events

within personality, or the self. We maintain that in adolescence, the person concept becomes

historical, and narrative becomes explicitly linked with the development of the self.

We have reviewed probable motivational and cognitive developmental precursors to the

emergence of the life story in detail elsewhere (Habermas & Bluck, 2000). In a general vein,

social-cognitive development in adolescence allows the construction of complex reflective

thinking (Fischer & Pruyne, 2003) and the complex combination of information from the

situation and from the individual’s history to predict behavior (Marini & Case, 1994). More

specifically, the development of the person concept in adolescence provides some support for the

development of the life story in adolescence. The person concept develops from an additive

model of unrelated habits to a unified psychological conception of personality in early

7

adolescence. In mid- to late adolescence a biographical dimension is added to the self-concept

(Guruccharri & Selman, 1982; Damon & Hart, 1986).

Narrative is a universal, basic form not only of remembering, but also of sharing and

understanding experiences. The narrative sequence follows the sequence of original events, and

narrative interprets and evaluates what has happened (Labov & Waletzky, 1967). Children learn

to narrate events in their preschool years and improve their skill during grade school years

(Peterson & McCabe, 1983). They learn to narrate in conversations with parents who implicitly

teach how to narrate by asking, complementing, and evaluating (Fivush, 2001). By the end of

grade school the ability to recount experiences is in place. Some childhood researchers seem to

suggest that narrative development or the development of a narrative identity reach their

endpoint in early (Stern, 1989) or late childhood (Nelson, 1998). However, only in early

adolescence does the ability to coordinate several episodes into a hierarchically organized, multi-

episode narrative (van den Broek, Lynch, & Naslund, 2003) emerge.

Two developmental studies of adolescent narrative development did show that adolescents

increasingly use biographical information to understand others. McKeough and Genereux (2003)

asked 10-, 12-, 14-, and 17-year-olds to write a story about someone with a problem and to

include flashbacks that provide a biographical background to the character. Flashbacks increased

in frequency with age, as did their interpretative quality. Feldman, Bruner, Kalmar, and Renderer

(1993) studied the comprehension of a short story by preadolescents, late adolescents, and adults.

Only the older adolescents and adults included biographical information about the protagonist to

explain and predict his actions.

A third developmental line relates more indirectly to the development of the life story. We

argue that constructing coherence in a life narrative requires complex interpretative abilities and,

8

correspondingly, the awareness that the facts of a life do not in themselves suffice to make a

convincing true story, but that a life needs to be interpreted and evaluated. There is pervasive

evidence that epistemological reasoning develops throughout adolescence, with an awareness of

the necessity of interpretation emerging only in mid-adolescence (King & Kitchener, 2004).

In addition to cognitive factors, the formation of a life story is probably motivated by

corresponding societal demands. These are reflected in biographical practices such as writing

diaries, collecting memorabilia, photographs and videos, reading biographies, and discussing

personal problems with close friends or parents.

Conceptual Framework and Research Strategy

To test the claim that the life story develops first in adolescence (McAdams, 1985), some

conceptual differentiations need to be introduced (cf. Habermas & Bluck, 2000). The life story is

defined by the coherence it creates among different circumstances and events in an individual

life, and between these and the development of the individual’s personality, including his or her

central values and life decisions. Like McAdams, we treat the term life story as the overarching

theoretical concept. McAdams’ idea of an internalized life story together with the literary theorist

Philippe Lejeune’s (1994) idea that everybody carries with him or her a raw version of her or his

life story that is continuously updated and revised, suggest a skeletal knowledge structure or life

story schema (Bluck & Habermas, 2000) containing the macrostructure of one’s life story. It may

be conceived of as the top level in a hierarchical model of the knowledge base of

autobiographical memory, which is activated whenever remembering or reflecting on the

personal past in the context of one’s life (Conway, Singer, & Tagini, 2004). While the life story

schema is a hypothetical construct, the life story manifests itself in two classes of phenomena.

Entire life narratives are linguistic products and are the most specific manifestation of the life

9

story. They may vary with the specific circumstances of their production, such as who they are

told to, for which purpose, and at which age. Still listeners expect life narratives to vary only

within the limits of historical truth (Conway, Singer, & Tagini, 2004). Listeners also expect life

narratives to be coherent and to explain how the narrator has become the person he or she is now.

Thus it is not enough to simply add narratives of specific incidents to each other without

connecting them, even if these are coherent within themselves. What is required is more than

local coherence between sentences or within single narratives. A life narrative requires global

coherence.

Life narratives are rarely produced in their entirety. The life story manifests itself most

frequently in a partial fashion in what we term autobiographical reasoning (Habermas & Bluck,

2000). It may show in arguments and narratives that establish links between distant parts of the

personal past and the personal present, between life experiences and the development of

personality.

In text linguistics, overall global coherence is understood to be “a complex, composite meta-

phenomenon, involving multiple strands” (Givon, 1995, p. 59). We have suggested that four

kinds of coherence are decisive for the overall global coherence of life narratives (Habermas &

Bluck, 2000). Global temporal coherence in life narratives refers to the listener’s orientation as

to at which point a given event has taken place in the narrator’s life. Global causal coherence in

life narratives refers to the listener’s orientation as to how actions and changes in personality are

caused and motivated by biographical experiences. Both temporal and causal coherence help the

listener to put events into sequence and to bridge change by providing temporal or causal-

motivational links. Global thematic coherence refers to the reader’s orientation as to which are

the dominant thematic strands in the life. Dominant themes create coherence by repeating and

10

varying the same theme again and again. A fourth kind of coherence is specific to life narratives,

which is coherence with a cultural concept of biography. The more a life narrative conforms to

cultural expectations of what a life story should look like, the more the listener is able to follow

the narrative. In this paper, we will be concerned with the first three kinds of textual coherence.

Conformity with the cultural concept of biography has also been studied with the sample used in

this study by comparing the seven most important memories selected for the life narratives with

a list of normatively salient biographical life events (Habermas, 2007; cf. Berntsen & Rubin,

2004).

The only other developmental study of free standing responses upon being asked for a life

narrative did not aim at measuring textual coherence but at describing the grammatical means

used to construct a responsible and active self, analyzing the use of modal verbs in a total of 36

five-, eight- and twelve-year olds (Quigley, 2000). However, Quigley only informally observes

that the 12-year olds’ anxiousness to conform to expected patterns of autobiography shows that it

is only at this age that children try to create a globally cohering autobiographical self in the strict

sense (Quigley, 2000, p. 158).

We argue that in order to study the development of a coherent life story, it is necessary to

construct measures of global coherence. We pursue two different strategies, one based on global

impressions of the listener or reader, the other based on formal elements of the text itself.

Listener-based ratings of aspects of global coherence do the greatest justice to the global nature

of coherence (cf. Fiese, Sameroff, Grotevant, Wamboldt, Dickstein & Fravel, 1999). Because life

narratives are complex texts, and coherence is a complex, multifaceted phenomenon, global

ratings are a very coarse measure that should be complemented by more objective, text-based

measures. Ratings of global coherence can only be applied to entire narratives, whereas local

11

indicators of global coherence may also be identified in more partial autobiographical texts such

as narratives of single memories. Here we only report measures used in whole life narratives,

while a separate test of autobiographical reasoning regarding one specific episode will be

reported elsewhere (Habermas, Fröhlich, & Diel, 2007).

To test the hypothesis that a coherent life story develops first in adolescence, we had to

develop an approach to autobiographical narratives that differs from other approaches because

our aim is to differentiate globally coherent from globally non-coherent (Linde, 1993) narratives.

First, we analyze entire life narratives, not just narratives of single memories. Second, we

complement ratings of the entire text with an analysis of formal syntactic-semantic elements.

Third, we assume that the contribution of these local elements to global coherence is additive, so

that relative frequency of indicators may be used as a continuous measure of degree of local

contributions to global coherence. Fourth, we do not interview participants by asking a series of

questions, but invite them to produce a free standing monologue in order to minimize the

influence of the interviewer on how much or little coherence is created. To compensate for the

absence of probing and countersuggestions, which are essential in the Piagetian tradition to test

for cognitive competence, several elements were introduced into the task and the design to

maximize participants’ production of coherence (see below). Fifth, oral and not written life

narratives were collected to lower the threshold for producing narratives and to eliminate writing

skill and ease as an additional factor contributing to the resulting narratives.

We briefly present the rationale for defining local indicators of global coherence and rating

scales and review the few studies in which they have been used. Almost all indicators used in

this study were developed in an earlier, exploratory study with a total of 12 12-, 15-, and 18-

12

year-olds and found to increase with age (Habermas &Paha, 2001). Indicators are grouped into

temporal, casal, and thematic indicators.

Temporal indicators that contribute to global temporal coherence are all those specifications

which allow the listener to locate events in the frame of the narrator’s life, such as specifications

of age, life phase, date, and distance to the present. Temporal specifications that relate only to

other parts of the narrative, such as ‘and then’ or ‘three days earlier’ do not by themselves

contribute to global temporal coherence in a life narrative.

Causal relations in a text express natural causes of events or human motives for actions and

evaluations. A mere count of causal relations, however, does not distinguish whether they

contribute to local or global coherence. We argue that if one of the causally linked statements

covers an extended period of time, this may indicate that the causal relation contributes to global

coherence. In a study of wisdom narratives of late adolescents, middle, and older adults (Bluck &

Glück, 2004), causal connections between an event and later life were named increasingly with

age.

Linde (1993) defined two more specifically biographical indicators of causal coherence in

life narratives, namely explanations of actions in terms of personality, and, vice versa,

explanations of (changes in) personality by events. Pasupathi found an increase in these

indicators across adulthood up to age 60 in some types of narratives (Pasupathi & Mansour,

2006; cf. Pasupathi, Mansour & Brubaker, 2007).

Causal coherence is also supported by biographical arguments which create causal

connections within the frame of life without requiring explicit causal syntactic links. Two

biographical arguments are learning a lesson for similar future situations and extracting an

abstract life maxim from an experience (McCabe, Capron, & Peterson, 1991; Pratt, Arnold,

13

Norris, & Filyer, 1999). McLean and Thorne (2003; McLean, 2005; Thorne, McLean, &

Lawrence, 2004) found either lessons or life maxims in 20% to 30% of narratives of single

memories by young adults. Bluck and Glück (2004) found an increase of lessons learned and

maxims mentioned in wisdom-related memories between adolescence and mid-adulthood.

Thematic coherence is inherently more content-based and thus harder to grasp for a formal

approach to coherence such as ours. Thematic coherence is created by relations of analogy or

similarity, such as those which are created by nuclear or prototypical episodes and their variants

(McAdams, 1985). These, however, tend to be highly individualistic. A more formal means to

create global coherence is found in exemplifications. In autobiographical narratives, general

statements are often backed up by specific episodes that illustrate them (Schütze, 1984), thereby

creating a hierarchical thematic link between general statement and specific instance.

Thematic coherence may be enhanced by the interpretive activity of the narrator. While the

development of a subjective perspective in narrative begins at preschool age with the inclusion of

evaluative statements (Fivush, 2001), only adolescents begin to realize the necessity of

interpreting past events by relating a variety of subjective perspectives and circumstances, as

described by studies of epistemological reasoning (King & Kitchener, 2004).

Although we do expect local indicators to contribute mostly to one kind of coherence, this

does not preclude that they also contribute to other kinds of coherence. For instance, temporal

order also offers information about implicit causal relations. In addition to the local indicators,

we constructed three rating scales to measure one aspect of each kind of global coherence for the

whole narrative. We expected local indicators contributing to a specific aspect of coherence to

correlate with the corresponding global rating.

14

As for the relative timing of the development of the three kinds of coherence, we expected

temporal coherence to develop first, followed by causal and then thematic coherence (Habermas

& Bluck, 2000). To measure ability and not just chance performance, participants need to be

enabled and motivated to narrate as coherently as they can. We therefore reduced the memory

load of the task, which is extraneous to the central ability to create coherence and differentially

affects participants of different ages. In addition, we asked for life narratives twice, two weeks

apart, giving one half of each age group a biographical training, while the control group was

given formally equivalent tasks with an unrelated content. We expected a combined effect of

mere repetition with training, so that the second life narratives of the training group could be

interpreted as reflecting ability more than the first narratives.

Three predictors of coherence other than age were also measured. Crystallized intelligence

has been found to be somewhat related to the quality of personal narratives (Pratt & Robins,

1991) and to the amount of free eyewitness recall, while there seems to be no relation to fluid

intelligence (cf. Henry & Gudjonsson, 2007). We explored the possible contribution of

crystallized and fluid intelligence to the telling of coherent life narratives. Secondly, it is

generally assumed that a major motivation to think about and create coherence in life is the

experience of unexpected events and turning points in life. This expectation is supported by the

finding that negative events are narrated and elaborated more extensively than positive ones

(Baumeister, Bratslavsky, Finkenauer & Vohs, 2001). We therefore asked participants to name

the number of negative life events experienced. Thirdly, creating coherence in life narratives

may be facilitated by everyday practices of discussing problems with others as well as by

biographical practices such as reading biographies or writing a diary.

15

This study tested the general hypothesis that the life story develops first during adolescence,

predicting an increase in a variety of indicators and ratings of temporal, causal, and thematic

coherence in life narratives. To ensure that we measured ability, participants were motivated to

perform at their maximum level by training and repeated measurement. Therefore we also

expected the maximum coherence to result in the second measurement after the training. The

main hypothesis was complemented by the second hypothesis that coherence was also predicted,

in addition to age, by intelligence, negative life events, and frequency of biographical practices.

Finally we tested the third hypothesis that measures of global coherence correlated with the

corresponding global rating.

Although this study is focused on general development, we will also test coherence for

gender effects, because there is some evidence that females narrate more, narrate more specific

events, and include more emotion words (Fivush, Brotman, Buckner, & Goodman, 2000;

Pillemer, Wink, DiDonao, & Sanborn, 2003), although gender differences are not always found

(Habermas, Ott, Schubert, Schneider, & Pate, in press).

Method

Participants

A total of 102 participants was divided into four groups with 25 eight-year-olds (M = 8.70,

SD = .24), 27 twelve-year-olds (M = 12.50, SD = .37), 25 sixteen-year-olds (M = 16.62, SD =

.45), and 25 twenty-year-olds (M = 20.57, SD = .49), almost equally divided between the sexes,

with 12, 14, 12, and 13 girls and young women included in the four age groups respectively. Six

other participants from all four age groups had to be excluded due to technical problems or

unwillingness to narrate.

The youngest were sampled from the top of the three classes of an elementary school in

16

Frankfurt am Main, since in Germany children are sent to different levels of school at age 10,

Gymnasium being the top level. The 12- and 16-year-olds were sampled from three classes of an

average Frankfurt secondary school, Gymnasium, comparable in the mixed social composition of

its students to the elementary school. The oldest age group was sampled from former students of

the same Gymnasium. The three younger groups were approached via their parents. We

presented the study at the parents’ meetings of each of the three classes in each age group, with

each class comprising between 24 and 32 students. Parents willing to participate returned written

informed consent by mail. The addresses of the oldest group were taken from published

‘graduation journals’ of the school, and former students were approached by mail. About 38% of

participants had at least one parent with a migrant background, ranging from East Asia to the

Americas, Southern and Eastern Europe and the Near East, distributed about equally between the

age groups. All participants spoke German fluently. Each participant received € 20.

Procedure

All participants were tested twice, two weeks apart, by two different (out of three) female

interviewers, who were distributed equally across conditions. In the first session they recounted a

life narrative. Roughly half of each age group (13, 14, 13, and 12 respectively in each age

group), divided about equally between the sexes, was trained by performing biography-related

tasks to enhance their ability to create a coherent life narrative. One such task was the

Autobiographical Reasoning Interview (Habermas & Fröhlich, 2007), in which one significant

event not told in the life narrative is asked to be related temporally, causally, and thematically to

other parts of life by posing specific questions. Two other tasks asked for the normative age of

25 life events and whether each of 40 life events were to be included or excluded from a life

narrative (Habermas, 2007). Here they were to be treated merely as training devices. The control

17

group was also divided roughly equally between age and sex, with a total of 49 participants.

They performed three tasks which were structurally similar to the training tasks, but focused on

neutral topics. Two weeks later, participants again recounted their life story and answered a

series of questionnaires. Sessions lasted between 45 and 60 minutes.

Material

Intelligence. Fluid intelligence was measured with the digit-symbol substitution subtest,

crystallized intelligence with the vocabulary subtest of the German version of the Wechsler

Intelligence Scale for Children – R (Tewes, 1991) or, for the 20-year-olds, with the German

version of the Wechsler Adult Intelligence Scales (Tewes, Rossmann & Schallberger, 1991). IQ-

values were higher in the group of 8-year-olds for both measures. Therefore we z-standardized

IQ scores separately for each age group.

Life events. Participants were asked whether they had ever experienced 14 specific life

events, taken from a German study of adolescents (Geyer, 1992a, b): death of parent, severe

illness, suicidal ideation, separation of parents, death of grandparent, loss of friend, parent job

losses, drop out of job training, birth of sibling, relocation, difficult final exam, leaving parental

home to go to school, change of school, and mandatory repetition of a school year.

Biographical practices. The frequency of five biographical practices (keeping a diary,

writing poems, looking at old pictures, reading old letters, reading biographies) and of confiding

in same- and cross-gender parent and friend (Zinnecker, 1985) were reported on a 9-point scale

ranging from very often to never. Biographical practices had a moderate reliability of α = .68,

confiding in others of α = .60.

Seven most important memories and life narratives. Participants were instructed twice, two

weeks apart, to recount their life stories for about 15 minutes. They were informed that the

18

second interviewer did not know the first life narrative. Prior to recounting their life, they wrote

their seven most important specific memories on index cards and ordered them in chronological

order on the table in front of them. They were instructed to include these seven memories in their

life narratives. The selection of the seven most important memories served two ends: We wanted

to make sure that all participants recounted specific events so that they had to integrate these into

their life narrative, and we wanted to prevent them from providing simply a very general

summary description of their life. Furthermore the cards served to reduce the memory load,

especially for the youngest age group.

Instructions said “First I’d like to ask you to think about the seven most important events

which happened in your life. These may be events that have just happened or they may have

happened a long time ago. Then please write your seven most important memories on these

seven cards. Please name only memories of very specific events. […] Now please arrange these

cards in the order in which they have happened on the table in front of you.” The following life

narrative instruction said: “Next I’d like you to tell me a story involving your whole life. Please

think about all the events that have happened in your life since you were born. Please integrate

the seven events into your story. For example you can tell me about the most important events in

your life and the biggest changes. You can tell me things which someone like me, who doesn’t

know you, might like to know about you. You can also tell me how what you have experienced

is still important to you today and how it has influenced what kind of person you are today.

Please take about 15 minutes time to tell your story. I will not interrupt you. After 10 minutes I

will tell you that you still have another five minutes. There are no right or wrong answers.”

Participants were asked to repeat the instruction to make sure they had understood it.

Interviewers did not intervene in the narrative, but encouraged participants to continue whenever

19

necessary. After the life narrative, participants dated all seven most important memories. The

additional eight participants told their lives without first being asked for important memories, so

as to explore the intended supportive effect of the memory cards for the younger children and to

exclude other distorting effects introduced by the memory cards.

Transcription, segmentation, and interrater reliabilities. Each life narrative was tape

recorded and transcribed verbatim. The text was divided into propositions. Propositions were

defined as all main or subordinate clauses that were understandable, even when incomplete, or as

a causal prepositional construction such as ‘because of …’ This allowed all causal relations to be

coded. Incomplete sentences that were not understandable were not counted as propositions. The

following example is the beginning of a life narrative of an eight-year old. Slashes separate

propositions:

“Well, first thing, I -, I was born / and then I grew bigger and bigger / and then was – my

first birthday with lots and lots of presents, and um many things / and then um my parents of

course had to work / so they sent me to Kindergarten and not to baby nursery / because –

she because a friend of my mother worked in kindergarten / and that is better / because

otherwise be- become come af– so afraid / …” (translated by the first author)

Two coders independently divided twenty life narratives balanced for sex, age, training vs.

control group, and measurement time into propositions, agreeing on 98.6% of the propositions.

The remaining life narratives were divided into propositions, half by each coder.

Then groups of three coders worked on seven groups of codes (temporal indicators, causal

relations and exemplifications, time span covered by statement, biographical arguments, hedges,

cognitive processes, comparisons between then and now), refining a manual that had been

successfully used in the aforementioned exploratory study (Habermas & Paha, 2001). The two

20

local indicators referring to causal links were constructed by combining the coding of causal

relations and the coding of the time span covered by the causally related statements, while the

other local indicators corresponded to the codes identified in the texts. Codes are categorical and

were coded for presence or absence in each proposition. Due to the dropout of one collaborator,

the interrater reliabilities could not be based on three coders for all codes. Therefore only four

out of the original seven Kappas are based on three coders (for these we report averaged

Kappas). Of the latter codes, we failed to reach a very good level of agreement for one code,

comparison between today and then, which remained around κ= .70. Rating scales were created

for this study (see below) and applied by two judges. All reliabilities are based on the

independent coding or rating of twenty life narratives balanced for age, gender, treatment, and

measurement time. Coders were initially blind to these characteristics, but narratives often

contained clues for age and gender. Coders were informed about the general thesis of the study,

but not about the hypotheses for single codes. After a good agreement of at least κ = .87 or ric =

.80 respectively was reached, one judge coded all the remaining narratives, while conflicting

codes for narratives used for exercise were resolved by discussion. To ensure that the one coder

did not deviate from the manual in the course of coding the remaining narratives, the other (one

or two) judge(s) coded eight extra narratives located at about one third and two thirds the way

through the remaining narratives to calculate agreement also in later phases of coding.

Most local indicators of global coherence were taken from an earlier, exploratory study of

the development of a coherent life story of 12-, 15- and 18-year olds (N = 12; Habermas & Paha,

2001). We present one group of indicators for temporal coherence, and three groups each for

causal and thematic coherence. The indicators of causal coherence include causally related

statements which cover an extended span of time, causal relations between statements about

21

events and personality, and a group of five biographical arguments. The indicators of thematic

coherence include exemplifications, specific hedges, and statements about specific complex

cognitive processes (see Table 1 for examples).

Indicators of temporal coherence. Several kinds of temporal indicators were identified

which allow the listener to locate an event in the life of the narrator (initial κ based on 20 life

narratives =.90, second κ based on two sets of four life narratives = .93).

Indicators of causal coherence. Two indicators of global coherence were constructed by

combining two codes. The first code was causal relations, the second the temporal extension of

causally related statements. Propositions were coded as causally related if they were causally

related in the strict sense (‘because’) or consecutive (‘so that’), contrastive (‘although’,

‘nevertheless’), conditional (‘if … then’), or instrumental (‘in order to’) following Sanders,

Spooren, and Noordman (1992, 1993; first κ =.88, second κ = .90.).

To distinguish between causal relations contributing to local and to global coherence, we

also coded the temporal extension of each causally related proposition. We coded circumscribed

events shorter than one month (‘and then he hit me, so that ...’), extended events lasting longer

than one month (‘in grade school I never fought with others, although ...’), and statements about

personality, i.e. about a personal state lasting longer than a month. This latter category of

personality included personality traits (‘I was someone who could never admit he was wrong’)

basic values (‘caring for my family is what is most important to me in my life’), and biographical

facts such as place of residence, profession, enduring physical attributes. We also coded

expressions of those personal states which are typically used by children to describe persons,

namely global social attitudes (‘nice’), preferences (‘I love spinach’), abilities (‘She was so good

in basketball’), and habits (‘I used to play soccer’; cf. Selman, 1980; κ = .87, second κ = .90).

22

Two related indicators of global causal coherence were constructed from causally related

statements. The first indicator ‘causally related statements of long duration’ identifies causally

related propositions concerning extended events or personality, thereby contributing to gloal

coherence. Secondly, we defined two codes for causal relations between statements about

personality and statements about events (of short or extended duration). Personality may be used

to explain an action, and events may be used to explain changes in personality.

In addition, five biographical arguments were coded. Reference to the developmental status

of a person may serve to explain something she or he did or was not able to do (McCabe,

Capron, & Peterson, 1991). Explaining a protagonist’s behavior by referring to his or her

individual biographical background and experiences is a second biographical argument

(Feldman, Bruner, Kalmar, & Renderer, 1993). Learning a lesson was coded when the narrator

said that she or he had learned to react in a certain way in a quite specific situation. The code is

comparable to ‘learning a lesson’ (McLean & Thorne, 2003) and ‘gaining factual or procedural

knowledge’ (Bluck & Glück, 2004). Life maxims were coded when generalizations to life

maxims or general rules of life were stated, which are comparable to ‘insight’ (McLean &

Thorne, 2003) and ‘life philosophy’ (Bluck & Glück, 2004). Finally we coded all explicit

statements about formative experiences and influences ( κ = .89, second κ = .84).

Indicators of thematic coherence. Exemplifications are narrations of specific events which

serve to render credibly to a general statement made earlier and were coded simultaneously with

causal relations. All hedges such as ‘maybe’ or ‘I’m not sure’ were coded, because in

combination with interpretative statements such as explanations or descriptions these express

awareness of an interpretative stance that motivates the construction of coherence. The objects of

the narrator’s insecurity were coded as explanation, evaluation, interpretation or description,

23

knowledge (‘He probably had a better job anyway’), memory (‘that was, I believe, in 3rd grade’),

and the process of narrating (‘I don’t know whether I’ve already mentioned …’; κ = .93, second

κ = .89). Only hedges regarding explanations and interpretations are used as indicators of global

coherence. Finally we coded statements of two complex cognitive processes which indicate the

subjective processing of events in the world, statements of opinion including convictions and

expectations lasting longer than at least a day, and statements of understanding or realizing

something (κ = .90, second κ = .89).

Global rating scales for temporal orientation, developmental consequentiality, and

plausible transitions. To complement the coding of the propositions, three 7-point scales were

used to rate global properties of whole life stories, related to temporal, causal, and thematic

coherence. Each was defined in a paragraph and by brief anchor definitions for each point (see

Table 1). Single intraclass correlations were computed between two raters on the basis of twenty

life narratives, resulting in correlations of ric = .80 (ric = .91 for the second correlation) regarding

temporal orientation, ric = .84 (ric = .91) for developmental consequences, and ric = .82 (ric = .85)

for plausible transitions. Remaining life narratives were rated half by each rater.

Results

Throughout the analyses, continuous variables were first tested for deviation from a normal

distribution and then for outliers. Outliers and variables were transformed to approach normal

distribution whenever necessary. For each of the three aspects of global coherence, we first

tested effects of age, training, and gender in multivariate analyses of variance (MANOVA) for

repeated measurements. Gender was controlled for because of the slightly unequal distribution

across age groups. Significant multivariate effects were followed up by univariate analyses.

Multivariate results are reported in the text. Means and standard deviations by age group are

24

listed in Table 2, together with F-values, significance and effect size for univariate main effects

of age, and with significant results of planned contrasts for differences between neighboring age

groups. Significant effects below p = .05 will be reported in the text for all other effects tested.

To limit the number of significance tests, only the seven summary local indicators (e.g., the sum

of all biographical arguments) and three ratings have been tested. In addition, means and

standard deviations of component local indicators (e.g., the single biographical arguments) are

reported for descriptive purposes. For each indicator the mean correlation between the two

measurement times will be reported. Age means are presented for all local indicators in Figure 1

and for the three ratings in Figure 2. Because some of the local indicators are, as was expected,

very rare in the youngest age group, the percentage of participants who use an indicator at least

once is presented in Table 3.

We will first test narratives for differences in length, then the main hypothesis of age effects

on global coherence in three MANOVAS, one for the indicators and ratings of each kind of

coherence. Then we will explore the correlations between ratings and local indicators. Finally we

will test additional predictors for their influence on global coherence.

Length of Life Narratives

The number of propositions (rt1,t2 = .77) increased significantly with age, as did, to a lesser

degree, the duration of narrations. Number of propositions differed significantly between ages 8

and 12 (p = .005) and ages 12 and 16 (p = .012), and likewise duration between ages 16 and 20

(p = .015). There was a trend for women to produce longer narratives (M = 229.7, SD = 79.0)

than men (M = 200.1, SD = 89.1), F(1, 86) = 3.87, p = .053, partial η2 = .04. The mean number

of words in propositions did not vary with age. To compensate for the differences in length of

narratives, we will compare not absolute but relative frequencies of indicators per proposition.

25

Effects of Age, Gender, Training and Retesting on Local Indicators and Ratings of Coherence

Temporal coherence. First, the sum of all indicators allowing to temporally locate an event

in the narrator’s life (rt1,t2 = .51) and the global rating of temporal orientation (rt1,t2 = .43) were

tested for effects of age, gender, training and retesting in a MANOVA. Significant multivariate

effects showed for age, Pillais’ F(6, 172) = 7.53, p = .000, partial η2 = .21, for the interaction of

age with gender, Pillais’ F(6, 172) = 2.39, p = .030, partial η2 = .08, and for retesting, Pillais’

F(2, 85) = 4.91, p = .010, partial η2 = .10. Univariate tests revealed the expected age effects in

the rating of temporal orientation, which increased monotonously with age (see Table 2 and

Figure 2), while local temporal indicators failed to reach significance (p = .072) when testing

overall differences between age groups (see also Figure 1). Inspection of age differences in

single temporal indicators (see Table 2) reveals that life phases were used increasingly with age

as were correct dates, which were near-absent in the youngest group (see Table 3). This suggests

that life narratives are increasingly anchored in calendar or historical time as well as in a

structured concept of the life course.

A significant interaction between age and gender showed in the rating of temporal

orientation, F(3, 86) = 2.99, p = .036, partial η2 = .09. Whereas temporal orientation increased

monotonously in women, men had lower values than women at ages 8 and 16, resulting in a local

maximum at age 12 and an absolute maximum at age 20. Finally, retesting resulted in an increase

in temporal orientation after two weeks, F(1, 86) = 8.97, p = .004, partial η2 = .09 (Mt1 = 4.68,

SD = 1.40, Mt2 = 5.17, SD = 1.57).

Causal coherence. To construct two of the local indicators of causal coherence, we

proceeded in two steps. First, all four kinds of causal relations were coded (causal-consecutive,

contrastive, conditional, instrumental; mean rt1,t2 = .29) which did increase almost monotonously

26

with age, F(3, 86) = 4.25, p = .001, partial η2 = .13. To differentiate contributions to global from

contributions to local causal coherence, we then constructed two specific indicators, namely the

relative frequency of causally related statements of long duration (mean rt1,t2 = .60) and causal

links between events (brief or extended) and personality (mean rt1,t2 = .19).

First causally related statements of long duration, causal relations between events and

personality, the sum of all biographical arguments (mean rt1,t2 = .32), and the global rating of

developmental consequentiality (mean rt1,t2 = .69) were tested for effects of age, gender, training

and retesting in a MANOVA. Significant multivariate effects showed for age, Pillais’ F(12, 255)

9.86, p = .000, partial η2 = .32, for retesting after two weeks, Pillais’ F(4, 83) = 3.46, p = .012,

partial η2 = .14, and for an interaction between retesting and age, Pillais’ F(12, 255) = 2.39, p =

.006, partial η2 = .10. Univariate tests revealed the expected age effects in all four dependent

variables.

It could be argued that the causal relations of long duration and causal relations between

events and personality only increased with age because all causal relations increased with age.

Therefore we also tested the effects of age on the frequency of these two indicators relative not

to all propositions but to all causal relations, with the same analysis of variance (ANOVA) (age,

gender, training, and repetition as factors). Both causal relations of long duration F(3, 86) =

43.79, p = .000, partial η2 = .60, and causal relations between events and personality continue to

increase significantly with age, F(3, 86) = 4.28, p = .010, partial η2 = .13 (with increasing age M

= 18.9%, SD = 12.4%, M = 28.7%, SD = 10.6%, M = 40.9%, SD = 10.9%, and M = 53.3%, SD =

10.8% of all causally related statements of long duration, and M = .01%, SD = .01% for age 8

and M = .02%, SD = .02% for each of the three older groups of all causal relations between

personality and events).

27

Since the distribution of relative frequencies of causal relations between personality and

events strongly deviated from a normal distribution due to the large number of participants who

did not use them even once (see Table 3), age differences were also tested non-parametrically

with the Jonckheere-Terpstra-test, which is designed to detect linear differences between ordered

classes such as age groups. Again there was a significant overall increase in the use of causal

relations between personality and events.

Univariate tests of the effects of retesting and of its interaction with age revealed significant

effects only for biographical arguments, with F(1, 86) = 11.47, p = .001, partial η2 = .12 for

retesting after two weeks, F(3, 86) = 7.04, p = .000, partial η2 = .20 for the interaction between

retesting with age. Inspection of means by age and measurement time showed that at ages 12 and

20 fewer biographical arguments were used at time 2 than at time 1.

Inspection of means (Table 2, Figures 1 and 2) reveals a monotonous increase with age in

all indicators except for causal relations between personality and events, for which an increase

was only measurable between the 8- and the 12-year olds, with no differences between the three

older age groups. Table 3 also shows that the causal relations between personality and events and

the biographical arguments are used by none or extremely few of the children in the youngest

age group, rising considerably in use up to age 20. This is strong evidence for the absence of

global coherence in life narratives in children.

Thematic coherence. Four dependent variables related to thematic coherence were tested for

the effects of age, gender, training and retesting in a MANOVA: the global rating of plausible

transitions (mean rt1,t2 = .22), exemplifications (mean rt1,t2 = .69), hedges (mean rt1,t2 = .59)

regarding explanations and descriptions, and cognitive processes of understanding and opinion

(mean rt1,t2 = .74). Significant multivariate effects showed for age, Pillais’ F(12, 255) = 6.56, p =

28

.000, partial η2 = .24, for gender, Pillais’ F(4, 83) = 3.82, p = .007, partial η2 = .16, and for an

interaction between retesting, age, and gender, Pillais’ F(12, 255) = 2.07, p = .019, partial η2 =

.09. Univariate tests revealed the expected monotonous age effects for all four dependent

variables (see Table 2, Figures 1 and 2).

Univariate tests revealed that female participants used more cognitive processes, F(1, 86) =

19.30, p = .000, partial η2 = .15 (M = 2.93%, SD = 1.64% versus M = 1.97%, SD = 1.59% in

men). A significant interaction between age, gender, and measurement time was found both for

exemplifications F(3, 86) = 2.94, p = .038, partial η2 = .09, and for ratings of plausible

transitions, F(3, 86) = 5.834, p = .026, partial η2 = .10. Inspection of means by age, gender, and

measurement time showed complex patterns. Deviations from a monotonous increase of

exemplifications with age showed for 16-year old males at the first testing and for 20-year old

females at the second testing, while deviations from a monotonous increase in plausible

transitions showed for 12-year old girls in the first testing and for 16-year old boys in the second

testing.

Only hedges regarding explanations and descriptions increased monotonously with age,

whereas hedges regarding knowledge decreased (Table 2), indicating that concerns regarding

correctness of statements, as reflected in the use of hedges, shifts with age from simple

knowledge to more interpretative descriptions and explanations.

Thus age had a pervasive effect on local indicators and ratings of global coherence.

Repetition had very few effects, namely an increase in the rated temporal orientation and a

decrease of the use of biographical arguments in two age groups, while neither training nor the

interaction of training and repetition had any effect whatsoever.

29

Intelligence, Life Events, and Biographical Practices.

We explored whether fluid and crystallized intelligence, number of negative life events,

biographical practices and confiding in others contribute to the prediction of coherence. Because

three of these predictors also correlated with age, number of life events r = .42, biographical

practices r = .27, and confiding in others r = .28, we calculated four hierarchical analyses of

covariance (ANCOVA) or multivariate analyses of covariance (MANCOVA), one for length of

narrative, and one for each kind of coherence. The factors age and gender and their interaction

were tested first, then the five continuous predictors were entered in the order listed above.

Dependent variables were averaged across measurement time. The rationale was to test whether

continuous variables could predict coherence in addition to age. Age was highly significant at the

p = .000 level in all analyses. Of the continuous predictors, only biographical practices

significantly predicted one kind of coherence, i.e., temporal coherence, F(2, 88) = 3.19, p = .046,

partial η2 = .07. However, in neither univariate analysis were biographical practices a significant

predictor. Crystallized intelligence reached borderline significance in the prediction of length of

life narratives, F(1, 89) = 3.90, p = .051, partial η2 = .04.

Intercorrelations between Local Indicators and Global Ratings

The text-based local indicators and global ratings were correlated with each other to test

whether indicators correlated with the ratings of the same aspects of coherence, encircled in

Table 4. We interpret correlations only with a significance of p < .01, that is from r = .26

upward, because of the great number of correlations (n = 21). Since most indicators correlated

highly with age, we partialled out age.

Of the local indicators, only temporal indicators, biographical arguments, and

exemplifications correlated with the corresponding global ratings. When testing the total

30

distribution of expected (3 significant and 13 non-significant) versus unexpected (1 significant

and 4 non-significant) correlations between local indicators on the one hand and the three global

ratings on the other hand, it proved to deviate significantly from chance, χ2(1, N = 21) = 3.86, p

= .049.

Discussion

This first study of the development of entire life narratives has shown an increase in the

relative frequency of a variety of local indicators and an increase in ratings of global coherence

across the adolescent age range, providing the first direct test of and overwhelming support for

McAdams’ (1985) hypothesis that the life story develops in adolescence.

Temporal coherence increases most between the ages 8 and 12, most clearly in the ratings

and with a similar trend in local indicators. This expected finding is mirrored by a comparable

developmental pattern in the acquisition of biographical knowledge in one half of the sample that

received training (Habermas, 2007). The near absence of calendar dates in the youngest age

group confirms experimental findings on the acquisition of calendar time between ages 10 and

12 (Friedman, 2004).

Both local indicators and global ratings of causal coherence increased with age as expected.

The relatively high percentage of participants in the three older age groups who used causal

event-personality links and biographical arguments at least once when compared with adults (cf.

Bluck & Glück, 2004; McLean & Thorme, 2005; Pasupathi & Monsour, 2006) probably results

both from the greater length of narratives analyzed here and from our wide definition of

personality, which included descriptions of personality typical for grade school children. This

wide definition was tailor-made for our age range. However, when analyzing biographical

31

reasoning, in the future an adult concept of personality should be used because only this

corresponds to an adult understanding of biography.

Thematic coherence also increased across all measures. Thus indicators and ratings of all

three kinds of global coherence increased with age, confirming the main hypothesis that global

coherence in life narratives develops across adolescence. However, it might be argued that the

hypothesis requires children to produce life narratives with a total lack of indicators of global

coherence. Some indicators indeed are missing in all or almost all of the youngest. These are also

the indicators that are most specifically biographical because they explicitly refer to personal

development, namely biographical arguments and events explaining personality development,

and also exemplifications and calendar dates. Knowledge of calendar time allows the most

precise temporal coordination of life events, and using exemplifications in response to the

demand to recount one’s life does reflect an understanding of the task as talking about events in

order to describe oneself. Thus the indicators which are most inherently biographical are nearly

missing in children, thereby providing even stronger support for McAdams’ (1985) hypothesis

than is given by a mere increase in indicators and ratings across adolescence.

Another objection to the interpretation of the results could be that the relative frequency of

central local indicators is low and that therefore their relevance is limited. However, in any given

life narrative the bulk of propositions is needed to provide the biographical material that is then

made coherent. It might even be more appropriate to compare the percentages of individuals in

each age group who do and don’t use, for example, a biographical argument (cf. Pasupathi &

Mansour, 2006). But even when pursuing a quantitative strategy, as was done in this paper, not

very many local indicators are required to render a life narrative coherent.

32

Inspection of global ratings suggests that temporal coherence increases most between ages 8

and 12, causal coherence most between ages 12 and 16, and thematic coherence most between

ages 16 and 20. Although this pattern does correspond to expectations (Habermas & Bluck,

2000), planned contrasts only partially confirm visual impressions. Moreover, the increases in

local indicators of causal and thematic coherence with age do show a different pattern, with the

steepest increase between ages 8 and 12.

Training and repetition of narrating one’s life showed few effects on global coherence.

Therefore we could not distinguish between mere performance and competence. On the other

hand, the stability of performance in spite of training suggests that limitations in coherence in

younger ages are not due to a lack of trivial skills or a lack of practice. To summarize, for the

first time this study provided direct evidence that the life story emerges and develops across

adolescence using the criterion of global textual coherence in life narratives.

In addition to age, neither intelligence, nor life events, nor biographical and confiding

practices predicted coherence. Possibly other cognitive and developmental factors discussed by

Habermas and Bluck (2000) influence the development of coherence in life narratives more than

the ones tested here.

Finally, correlations between local indicators and ratings only partially conformed to the

hypothesized pattern. For each kind of coherence, one local indicator correlated with the correct

rating scale. These were temporal indicators, biographical arguments, and exemplifications, that

is, three of the four indicators that had very low frequencies in the youngest group. Therefore

concurrent evidence points to these local indicators as specific for the respective kind of

coherence. In addition, if personality was defined more narrowly, causal relations between

events and personality might also qualify as a good indicator of causal coherence. Possibly

33

causally related statements of long duration are not specific enough to life narratives, and hedges

regarding explanations and interpretations and complex cognitive processes are too indirectly

related to global thematic coherence to be good indicators.

A major limitation of this study is that it was restricted to a relatively homogeneous sample in

terms of educational level and cultural identity. Although the cultural background of

participants’ parents was quite diverse, as is typical for a major European city, the content of life

narratives suggests that almost all participants were well immersed in German culture. Moreover,

the two older groups showed a very good knowledge of cultural age norms (Habermas, 2007).

Anthropological evidence points to the non-universality of telling one’s life (Tonkin, 1992) and

developmental evidence points to strong cultural differences in remembering and the related

construction of the self (Wang, 2004). As a consequence, possible variations in the development

of life narratives remain to be studied both in educationally and culturally distinct groups. In

addition, cohort effects cannot be excluded. However, the maximal difference in birth year is

only 12 years, rendering cohort differences improbable.

Future research needs to address three areas of interest. First, the attempt to quantify a

complex phenomenon such as the global coherence in life narratives warrants the development

of additional, qualitative measures especially for thematic coherence, such as explicit statements

of recurrent themes in life or metaphors of the self that integrate a variety of experiences. Also,

for temporal coherence deviations from a linear temporal order could simply be counted. Global

ratings of coherence could be supplied by experts such as teachers of literature. These additional

measures should correlate with the more successful measures introduced here.

Also, the development of the life story in adolescence should be compared to related

phenomena. The expected close relationship between life narrative coherence and

34

autobiographical reasoning in more partial narratives, for example, could be confirmed in half of

this sample (Habermas, Fröhlich, & Diel, 2007). The development of coherence in life narratives

should also be demonstrated in the spontaneous tendency to use the past to define the current

self, for example, in free self-descriptions. Our theoretical assumptions also predict that

autobiographical reasoning parallels what could be called heterobiographical reasoning, that is

the ability to create biographical coherence in narratives of others’ lives. Other related constructs

differ from the life story due to their focus on individual differences, such as Marcia’s (1966)

identity status and attachment security. The moderate positive correlation between ratings of

‘meaning making’ in turning point narratives, which corresponds to our indicators ‘gaining

insight’ and ‘life maxims’, and an index supposedly reflecting identity achievement (McLean &

Pratt, 2006) does not contradict a more distant relationship between the constructs. Similarly,

although coherence is the single most important criterion for attachment security in the Adult

Attachment Interview (AAI; Main, Kaplan, & Cassidy, 1985), this kind of coherence is less

formal and more content-based, asking for the absence of incoherence or implicit contradictions.

We suggest that the relation between the life story and coherence in the AAI is one of

presupposition: Answering the questions of the AAI directed at a biographical understanding of

one’s life requires the ability for autobiographical reasoning. Accordingly, the AAI is not

administered before about age 15.

Finally, there is a need to explore social, social-cognitive, and motivational developments

that possibly facilitate the development of the life story. Neither the number of negative life

events nor the frequency of biographical practices were related to the life story in this study.

Social practices such as specific ways of talking about lives with parents, similar to the

35

socializing practices of memory talk in preschool children (Haden, Hayne, & Fivush, 1997),

might be influential, as might more general social-cognitive developments in adolescence.

36

References

Allport, G. W. (1942). The use of personal documents in psychological science. New York:

Social Science Research Council.

Antonovsky, A. (1985). The life cycle, mental health and the sense of coherence. Israel Journal

of Psychiatry and related Sciences, 22, 72-80.

Bauer, P. J. (2006). Remembering the times of our lives: Memory in infancy and beyond.

Mahwah, CT: Erlbaum.

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Voihs, K. D. (2001). Bad is stronger than

good. Review of General Psychology, 5, 323-370.

Berntsen, D., & Rubin, D. C. (2004). Cultural life scripts structure recall from autobiographical

memory. Memory & Cognition, 32, 427-442.

Bluck, S., & Glück, J. (2004). Making things better and learning a lesson: Experiencing wisdom

across the life span. Journal of Personality, 72, 543-572.

Bluck, S., & Habermas, T. (2000). The life story schema. Motivation & Emotion, 24, 121-147.

Conway, M., Singer, A., & Tagini, A. (2004). The self and autobiographical memory:

Correspondence and coherence. Social Cognition, 22, 491-529.

Damon, W., & Hart, D. (1986). Stability and change in children's self-understanding. Social

Cognition, 4, 102-118.

Demetriou, A. (2000). Organization and development of self-understanding and self-regulation:

Towards a general theory. In M. Boekaerts, P. R. Pintrich & M. Zeidner (Eds.), Handbook of

self-regulation (pp. 209-251). San Diego, CA: Academic Press.

Erikson, E. H. (1968). Identity, youth and crisis. New York: Norton.

Feldman, C., Bruner, J., Kalmar, D., & Renderer, B. (1993). Plot, plight, and dramatism:

Interpretation at three ages. Human Development, 36, 327-342.

37

Fiese, B. H., Sameroff, A. J., Grotevant, H. D., Wamboldt, F. S., Dickstein, S., & Fravel, D. L.

(1999). The stories that families tell: Narrative coherence, narrative interaction, and

relationship beliefs. Monographs of the Society for Research in Child Development, 64(2,

serial No. 257).

Fischer, K., & Pruyne, E. (2003). Reflective thinking in adulthood: Emergence, development,

and variation. In J. Demick & C. Andreoletti (Eds.), Handbook of adult development (pp.

169-198). New York: Kluwer Academic/Plenum Publishers.

Fivush, R. (2001). Owning experience: Developing subjective perspective in autobiographical

narratives. In C. Moore & K. Lemmon (Eds.), The self in time: Developmental perspectives

(pp. 35-52). Mahwah, NJ: Erlbaum.

Fivush, R., Brotman, M. A., Buckner, J. P., & Goodman, S. H. 2000). Gender differences in

parent –child emotion narratives. Sex Roles, 42, 233-253.

Freud, S. (1905). Fragment of an analysis of a case of hysteria. In S. Freud (1975). Standard

Edition, Vol. 7 (pp. 7–122). London: Hogarth Press.

Friedman, W. J. (2004). Time in autobiographical memory. Social Cognition, 22, 591-605.

Geyer, S. (1992). Lebensverändernde Ereignisse, soziale Ungleichheit und Belastungen [Life-

changing events, social inequality, and stress]. In J. Zinnecker (Ed.), Jugend ´92, Band 2 (p.

221 - 237). Opladen, GE: Leske & Budrich. (a)

Geyer, S. (1992). Methodische Aspekte der Erfassung lebensverändernder Ereignisse

[Methodological aspects of measuring life-changing events]. In J. Zinnecker (Ed.), Jugend

´92, Band 4 (p.27-40). Opladen, GE: Leske & Budrich. (b)

Givón, T. (1995). Coherence in text vs. coherence in mind. In A. Gernsbacher & T. Givón

(Eds.). Coherence in spontaneous text (pp. 59-115). Amsterdam: Benjamins.

38

Gregg, G. S. (2006). The raw and the bland: A structural model of narrative identity. In D. P.

McAdams, R. Josselson, & A. Lieblich (Eds.), Identity and story: Creating self in narrative

(pp. 63-87). Washington DC: American Psychological Association.

Gurucharri, C., & Selman, R. (1982). The development of interpersonal understanding during

childhood, preadolescence, and adolescence. Child Development, 53, 924-927.

Habermas, T. (2007). How to tell a life: The development of the cultural concept of biography

across the lifespan. Journal of Cognition and Development, 8, 1-31.

Habermas, T., & Bluck, S. (2000). Getting a life: The development of the life story in

adolescence. Psychological Bulletin, 126, 748-769.

Habermas, T., & Paha, C. (2001). The development of coherence in adolescents’ life narratives.

Narrative Inquiry, 11, 35-54.

Habermas, T., Fröhlich, A., & Diel, V. (2007). Autobiographical reasoning from late childhood

to early adulthood: The Autobiographical Reasoning Interview. Unpublished manuscript,

University of Frankfurt am Main.

Habermas, T., Ott, L. M., Schubert, M., Schneider, B., & Pate, A. (in press). Stuck in the past:

Negative bias. explanatory style, temporal order, and evaluative perspectives in life

narratives of clinically depressed individuals. Depression and Anxiety.

Haden, C. A, Haine, R. A., & Fivush, R. (1997). Developing narrative structure in parent-child

reminiscing across the preschool years. Developmental Psychology 33, 295-307.

Henry, L. A., & Gudjonsson, G. H. (2007). Individual and developmental differences in

eyewitness recall and suggestibility in children with intellectual disabilities. Applied

Cognitive Psychology, 21, 361-381.

Kernberg, O. F. (1984). Severe personality disorders. New Heaven, CT: Yale University Press.

39

King, P. M., & Kitchener, K. S. (2004). Reflective judgment: Theory and research on the

development of epistemic assumptions through adulthood. Educational Psychologist, 39, 5-

18.

Kroger, J. (2007). Identity development: Adolescence through adulthood. Thousand Oaks, CA:

Sage.

Labov, W., & Waletzky, J. (1967). Narrative analysis: Oral versions of personal experience. In I.

Helm (Ed. ), Essays on the verbal and visual arts. Proceedings of the 1966 Annual Spring

Meeting of the American Ethnological Society (pp. 12-44). Seattle: University of Washington

Press.

Lejeune, P. (1975). On autobiography. Minneapolis: University of Minnesota Press.

Linde, C. (1993). Life stories. Cambridge, UK: Cambridge University Press.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and attachment: A

move to the level of representation. In I. Bretherton, & E. Waters (Eds. ), Growing points of

attachment research. Monographs of the Society for Research in Child Development, 50(1-2),

66-104.

Marcia, J. E. (1966). Development and validation of ego identity status. Journal of Personality

and Social Psychology, 3, 551-558.

Marini, Z., & Case, R. (1994). The development of abstract reasoning about the physical and

social world. Child Development, 65, 147-159.

McAdams, D. P. (1985). Power and Intimacy. New York: Guilford.

McAdams, D. P., & Pals, J. (2006). A new big five: Fundamental principles for an integrative

science of personality. American Psychologist, 61, 204-217.

40

McAdams, D. P., Bauer, J. J., Sakaeda, A. R., Anyidoho, N. A., Machado, M. A., Magrino-

Failla, et al. (2006). Continuity and change in the life story: A longitudinal study of

autobiographical memories in emerging adulthood. Journal of Personality, 74, 1371-1400.

McCabe, A., Capron, E., & Peterson, C. (1991). The voice of experience: The recall of early

childhood and adolescent memories by young adults. In A. McCabe & C. Peterson (Eds.),

Developing narrative structure (pp. 137-173). Hillsdale, NJ: Erlbaum.

McKeough, A., & Genereux, R. (2003). Transformation in narrative thought during adolescence:

The structure and content of story composition. Journal of Educational Psychology, 95, 537-

557.

McLean, K. (2005). Late adolescent identity development: Narrative meaning making and

memory telling. Developmental Psychology, 41, 683-691.

McLean, K., & Pratt, M. W. (2006). Life’s little (and big) lessons: identity status and meaning

making in turning point narratives of emerging adults. Developmental Psychology, 42, 714-

722.

McLean, K., & Thorne, A. (2003). Late adolescents’ self defining memories about relationships.

Developmental Psychology, 30, 635-645.

Nelson, K. (1998). Language in cognitive development – The emergence of the mediated mind.

Cambridge, UK: Cambridge University Press.

Nelson, K., & Fivush, R. (2004). The emergence of autobiographical memory: A social cultural

developmental theory. Psychological Review. 111, 486-511

Pals, J. L. (2006). Constructing the "Springboard Effect": Causal Connections, Self-Making, and

Growth Within the Life Story. In D. P. McAdams, R. Josselson, & A. Lieblich, (Eds), Identity

41

and story: Creating self in narrative (pp. 175-199). Washington DC: American Psychological

Association.

Pasupathi, M., & Mansour, E. (2006). Adult Age Differences in Autobiographical Reasoning in

Narrative. Developmental Psychology, 42, 798-808.

Pasupathi, M., Mansour, E., & Brubaker, J. R. (2007). Developing a Life Story: Constructing

Relations between Self and Experience in Autobiographical Narratives. Human Development,

50, 85-110.

Peterson, C., & McCabe, A. (1983). Developmental psycholinguistics: Three ways of looking at

a child's narrative. New York: Plenum.

Pillemer, D. B., Wink, P., DiDonato, T. E., & Sanborn, R. L. (2003). Gender differences in

autobiographical memory styles of older adults. Memory, 11, 525-532.

Pratt, M. W., Arnold, M. L., Norris, J. E., & Filyer, R. (1999). Generativity and moral

development as predictors of value socialization narratives for young persons across the adult

life span: From lessons learned to stories shared. Psychology and Aging, 14, 414-426.

Pratt, M. W., & Robins, S. L. (1991). That’s the way i twas : Age differences in the structure and

quality of adults’ personal narratives. Discourse Processes, 14, 73-85.

Quigley, J. (2000). The grammar of autobiography – A developmental account. Mawah, NJ:

Erlbaum.

Rubin, D. C., Rahal, T. A., & Poon, L. W. (1998). Things learned in early adulthood are

remembered best. Memory and Cognition, 26, 3-1.

Sanders, T. J. M., Spooren, W. P. M., & Noordman, L. G. M. (1992). Toward a taxonomy of

coherence relations. Discourse Processes, 15, 1-35.

Sanders, T. J. M., Spooren, W. P. M., & Noordman, L. G. M. (1993). Coherence relations in a

42

cognitive theory of discourse representation. Cognitive Linguistics, 4, 93-133.

Schrauf, R. W., & Rubin, D. (2001). Effects of voluntary immigration on the distribution of

autobiographical memories over the lifespan. Applied Cognitive Psychology, 15, S75-S88.

Schütze, F. (1984). Kognitive Figuren des autobiographischen Stegreiferzählens [Cognitive

structures of improvised personal narratives]. In M. Kohli & G. Robert (Eds.), Biographie

und soziale Wirklichkeit (pp. 78-118). Stuttgart, GE: Metzler.

Selman, R. (1980). The development of social understanding. New York: Academic Press.

Singer, J. A. (2004). Narrative identity and meaning making across the adult lifespan: An

introduction. Journal of Personality, 72, 437-460.

Stem, D. N. (1989). Developmental prerequisites for the sense of a narrated self. In A. M.

Cooper, O. F. Kemberg, & E. S. Person (Eds.), Psychoanalysis: Toward the second century

(pp. 168-178). New Haven, CT: Yale University Press.

Tewes, U. (1991). HAWIE-R. Bern, SW: Huber.

Tewes, U., Rossmann, P., & Schallberger, U. (1991). HAWIK-III. Bern, SW: Huber.

Thorne, A., McLean, K., & Lawrence, A.M. (2004). When remembering is not enough:

reflecting on self-defining memories in late adolescence. Journal of Personality, 72, 513-542.

Tonkin, E. (1992). Narrating our past – The social construction of oral history. Cambridge. UK:

Cambridge University Press.

Van den Broek, P., Lynch, J., & Naslund, J. (2003). The development of comprehension of main

ideas in narratives: Evidence from the selection of titles. Journal of Educational

Psychology, 95 (4), 707-18.

43

van Hoof, A. (1999). The identity status field re-reviewed: An update of unresolved and

neglected issues with a view on some alternative approaches. Developmental Review, 19,

497-556.

Wang, Q. (2004). The emergence of cultural self-constructs: Autobiographical memory and self-

description in European American and Chinese children. Developmental Psychology, 40, 3-

15.

Zinnecker, J. (1985). Literarische und ästhetische Praxen in Jugendkultur und Jugendbiographie

[Literary and aesthetic practices in youth culture and youth biography]. In A. Fischer, W.

Fuchs & J. Zinnecker (Hg.), Jugendliche und Erwachsene 1985, Bd. 2 (pp. 143 - 348).

Opladen, GE: Leske & Budrich.

44

Table 1

Examples of Local Indicators of Global Coherence Identified at the Level of Propositions and

Definitions of Global Ratings

Code Example

__________________________________________________________________________

Local indicators of temporal global coherence

1. Temporal indicators

Life phase “I got to know her in fourth grade”

“When I was a baby”

Age “When I was 9”

Calendar date “in 2002” or “on May 6, 2006”

Distance from present “that was last year”, “two years ago”

Local indicators of causal global coherence

2. Causally related statements of long duration<L:

Extended event “I would really like to have a real change in my life, because

everything’s gotten so confused.”

Personality “I was a very shy person then. Therefore I never dared to ask her”

3. Causal links between personal states and events

Personality “We did a lot of boating, because my father likes rowing.”

explains action “I actually had problems with the teachers throughout school,

because I am someone who doesn’t like to adapt himself.”

Event explains “If I hadn’t joined the sports-club then, I wouldn’t have friends

personality now and I would still be a shy person.”

“After age 10 I became a shy person because the separation of my

parents made me distrust others”

4. Biographical arguments

Developmental status “My parents' divorce didn’t affect me much. I wasn’t really aware

of what was happening - I was still too little.”

Biographical background “When suddenly a car raced towards us, he panicked. He had been

45

run over by a car when he was small”

“All throughout childhood my parents helped me with everything,

they took care of all the paperwork, everything. And now I am 18

and I am still unable to organize my life.”

Lesson learned “That’s why I told myself, next time I fall in love, school work

should not suffer from it.”

“I swallowed a cent. Then I learned that I better not swallow small

objects”

Life maxim, generalization “I had received a bad mark, and I thought that she had given marks

in a mean way. That’s what you usually think when you get a bad

mark “

“I was really emotionally hooked up with him for a long time.

Probably that’s what always happens when it’s the first kiss.”

Formative experience “My first relationship has influenced me extremely.”

Local indicators of thematic global coherence

5. Illustration “I haven’t learned to do things on my own [general statement],

although I am able to use the subway on my own and stuff like that,

but, for example, I do not have the faintest idea how to go about

anything bureaucratic. [exemplification]”

6. Hedges

Regarding explanations “Anna is a very complicated person, just like me. Maybe that’s why

we make such a good fit.”

“Maybe that’s why I am so lazy”

Regarding descriptions or “I’m not sure whether I’m going to be good at the exam”

interpretations “Now I am better able to control myself, I believe”

7. Complex cognitive processes

Understanding “Even then I had understood that my mother was mentally ill”

“Then I realized, how little I meant to him”

“Then I had already understood that my mother was mentally ill”

46

Opinion “I was always convinced that he didn’t care for me”

“I was of the opinion that I didn’t need it”

Global rating scales (Values range from 1 to 7 points)

8. Rating of temporal orientation (temporal coherence)

Degree to which the reader is able to follow the temporal sequence or/and the absolute timing of

events in life, ranging from 0 to 100% of events.

9. Rating of developmental consequentiality (causal coherence)

Degree to which the reader understands how past experiences explain how personality, life, or

outlook have changed and what the turning points and motives for these developments were,

ranging from ‘no change in personality is described’ to ‘the development of the personality is

presented with its turning points and motives.’

10. Rating of plausible transitions (thematic coherence)

Degree to which events followed each other in an unrelated way versus the degree to which

implicit or explicit plausible links between events were provided. Rater had to decide on the

relative amount of events or themes that had good transitions to the following event or theme,

again ranging from 0 to 100%.

Table 2

Means and Standard Deviations of Length of Life Narratives, Relative Frequencies of Local Indicators, and of Global Ratings

Age Group ANOVA Age Groupa

8 12_ 16 20_ partial significant

M SD M SD M SD M SD F(3,86) η2 contrastsb

Number of propositions 146.66 68.37 201.04 86.57 248.58 88.93 274.80 63.34 17.01*** .37 8 < 12 < 16

Duration (minutes) 11.95 3.14 12.54 2.60 13.47 2.33 14.49 2.64 5.14** .15 16 < 20

Percent of propositions

1. All temporal indicators 3.79 2.01 5.09 3.09 5.23 1.52 6.30 1.95 2.42 .08

Life phase 1.93 1.32 1.91 1.15 2.58 1.25 3.08 1.39

Age 1.37 1.08 1.72 1.63 1.13 .82 1.50 .78

Calenda date .02 .06 .74 1.56 .37 .48 .94 1.11

Distance from present .48 .56 .73 .59 1.15 .72 .78 .61

All causal relations 7.84 3.61 9.67 3.01 10.59 2.48 10.29 2.53

2. Causal rel. long duration 1.34 .83 2.85 1.38 4.38 1.45 5.51 1.74 44.20*** .61 8 < 12 < 16 < 20

3. Causal rel. pers. + events .11 .24 .51 .55 .49 .42 .51 .53 7.24*** .20 8 < 12

Personality explains event .09 .24 .33 .41 .32 .33 .34 .38

Event explains personality .02 .07 .17 .23 .17 .25 .17 .26

4. All biographical arguments .32 .52 .55 .55 1.08 .90 1.48 1.18 14.99*** .39 12 < 16

Developmental status .08 .19 .20 .24 .25 .26 .27 .30

Biographical background .01 .04 .11 .28 .15 .26 .06 .10

Lesson learned .00 .00 .01 .06 .05 .12 .00 .00

Life maxim .23 .51 .23 .32 .44 .54 .69 .61

Formative experience .00 .00 .00 .00 .20 .45 .46 .51

5. Exemplification .18 .24 .56 .45 .64 .50 .92 .51 20.77*** .42 8 < 12, 16 < 20

6. Hedges explan. interpret. 1.15 .85 2.61 1.86 2.89 1.51 3.55 1.70 10.17*** .26 8 < 12

Explanation .26 .29 .54 .55 .58 .59 .67 .50

Interpretation .89 .80 2.12 1.57 2.31 1.15 2.88 1.34

Other hedges regarding

Knowledge .35 .37 .31 .32 .16 .19 .10 .12

Memory .99 1.05 .69 .77 .83 .71 .76 .72

Narration .01 .02 .14 .18 .39 .41 .31 .29

7. Complex cognitive proc. .87 .84 2.35 1.71 2.93 1.62 3.81 1.23 28.81*** .50 8 < 12, 16 < 20

Understanding .23 .37 .43 .34 .70 .88 .98 .61

Opinion .64 .77 1.92 1.53 2.24 1.28 2.84 1.25

Rating scales (values 1 to 7)

8. Temporal orientation 3.76 1.36 5.02 1.09 5.16 .87 5.90 .76 19.77*** .41 8 < 12, 16 < 20

9. Developmental consequent. 1.52 .37 2.23 .73 4.08 1.26 4.80 1.22 69.74*** .69 8 < 12 < 16 < 20

10. Plausible transitions 2.76 1.01 2.98 1.14 3.10 .92 3.70 .94 4.00** .12 12 < 16 < 20

Additional predictors

Number of life events 2.84 1.28 4.19 1.69 3.47 1.95 5.61 1.96

Autobiographical practices 1.65 .33 1.89 .31 1.92 .34 1.91 .32

Confiding in others 3.14 1.97 3.68 1.08 4.11 1.69 4.40 1.64

Note. Only the seven local indicators have been tested for significance. Means and standard deviations of the component codes of local

indicators and of some other variables for which no hypotheses were formulated are set in italics and are reported merely for

descriptive purposes.

aOnly age effects are reported here, significant effects of gender, treatment and measurement time and all interactions are reported in

the results section.

b Planned contrasts were to test the differences between neighboring age groups. Significant differences are indicated by < or >, no

significant difference by a =.

Table 3

Percentage of Participants Who Use Local Indicators at least once

in any of the Two Life Narratives

Age group

Local indicator 8 12 16 20

1. All temporal indicators 100.0 100.0 100.0 100.0

Life phase 92.0 100.0 96.0 100.0

Age 92.0 81.5 92.5 100.0

Date 8.0 59.3 56.0 92.0

Distance from present 56.0 85.2 96.0 88.0

2. Causal long duration 96.0 100.0 100.0 100.0

3. Causal event-personality 20.0 77.8 76.0 80.0

Personality expl. event 12.0 66.7 72.0 68.0

Event expl. personality 8.0 54.4 54.0 40.0

4. All biogr. arguments 44.0 74.1 92.0 100.0

Developmental status 20.0 51.9 68.0 64.0

Biograph. background 4.0 22.2 40.0 32.0

Lessons learned 0 3.7 16.0 0

Life maxim 24.0 40.7 68. 0 88.0

Formative experience 0 0 28.0 68.0

5. Exemplifications 44.0 81.5 88.0 96.0

6. Hedges 80.0 85.2 100.0 100.0

regarding explanations 52.0 66.7 88.0 84.0

regarding interpretation 80.0 85.2 100.0 100.0

7. Complex cognitive proc. 68.0 92.6 100.0 100.0

Understanding 36.0 77.8 80.0 92.0

Opinion 56.0 92.6 100.0 100.0

Table 4

Correlations of narrative coherence variables with age and partial correlations among

coherence variables, age partialled out

Measure Age 1 2 3 4 5 6 7 8 9

Local Indicators

1. Temporal indicators .24

2. Long durationa .75 -.07

3. Events-personalityb .37 -.01 .40

4. Biographical arg. .52 -.02 .08 .20

5. Exemplifications .53 -.22 .00 .43 .08

6. Hedges explan. interpr.c .48 -.07 .08 -.03 -.11 .01

7. Cognitive processes .60 -.16 .09 .11 .19 .09 .07

Rating Scales

8. Temporal Orientation .57 .35 -.04 .07 .00 .05 -.11 .06

9. Dev. consequentiality .80 -.12 .13 .14 .39 .14 -.05 .22 -.06

10. Plausible transitions .31 -.29 .08 .10 .01 .26 .03 .02 .17 .11

Note. Expected significant correlations between indicators of the same aspect of global

coherence are circled.

aCausally related propositions about states lasting longer than one month bCausal

relations between personality and events. cHedges regarding explanations and descriptions

and interpretations.

Levels of significance for correlations p < .01 for r = .26, p < .001 for r = .32. all p

<.05 are marked in bold.

Figure Captions

Figure 1. Mean percentage of propositions with local indicators of global coherence - temporal

indicators, causally related propositions of long duration, hedges regarding interpretations and

descriptions, and cognitive processes regarding understanding and opinion by age, averaged

across measurement times.

Figure 2. Mean percentage of propositions with local indicators of global coherence -

biographical arguments, exemplifications, and causal relations between events and personality by

age, averaged across measurement times.

Figure 3. Mean values on global ratings for temporal orientation, plausible transitions between

events, and developmental consequentiality by age, averaged across measurement times.

55

8 12 16 20

Age

0

1

2

3

4

5

6

7M

ean

% o

f Pro

posi

tions

Temporal IndicatorsLong DurationHedgesCognitive Processes

8 12 16 20

Age

0,0

0,5

1,0

1,5

Mea

n %

of P

ropo

sitio

ns

Biograph. ArgumentsExemplifi-cationsEvent-Personality

56

8 12 16 20Age

1

2

3

4

5

6 Temporal OrientationReasonable TransitionsDevelop-mental Conse-quentiality